← BACK TO HOME

Search, Discovery, and Getting It Right Before Launch.

Launch was approaching. The UChicagoNode platform didn't feel ready, but instinct alone isn't enough to delay a launch date.

A pre-launch study paused the launch, and reshaped the roadmap.

feeling isn't evidence — now what?

TIMELINE:

Feb – Apr 2026

ROLE:

UX Designer: Sole Design Contributor

TEAM:

5 Metadata Specialists, 1 PM, 2 Developers, Director

OUTCOME:

Critical issues identified and resolved.

Context:

A pre-launch study that reshaped the roadmap.

UChicagoNode is a unified digital collections platform for the University of Chicago Library. With launch approaching, something felt off, the platform didn't feel ready for the 120,000+ researchers, librarians, and students who would use it. But instinct alone wasn't enough to push back on a launch date. Evidence was. That's why I proposed a targeted pre-launch study, to find out if what I was sensing was actually there.

UChicagoNode digital collections platform.

Problem:

what was the problem?

1. Unreliable Search:
Singular vs. plural queries returned different results. Multi-word searches returned zero.
2. Advanced Search Confusion:
Too many options upfront, submit button buried below a long scroll.
3. Performance:
Slow load times — one search took 29 seconds to execute.

Process:

1. survey.

designed and shipped in under a week.

A 7-section Google Form sent to library partners: metadata specialists, GIS librarians, and digital collections managers, undergraduate students: the people who would live in the platform daily.

25+ responses  |  2 weeks from design to synthesis  |  7 sections

Google Form showing Section 2 of 7 of the UChicagoNode pre-launch survey
The 7-section survey form sent to library partners.

2. what they said.

what did the survey reveal?

"With Advanced Search, it was hard to find the 'submit query' button, all the way at the bottom of the page."

— Survey Respondent

"If I search 'Pilsen map' or 'Pilsen AND map', there are 0 results. Individual terms work fine."

— Survey Respondent

"Speed is a real issue — it took 29 seconds to execute a search for 'Chicago.'"

— Survey Respondent

"Is there an explanation for Open Access? Not all users may understand the distinction between open access vs permission to use."

— Survey Respondent

3. synthesis.

10 themes from FigJam affinity mapping.

Search Retrieval Search Ranking Advanced Search Confusion Filter / Facet Reliability Metadata Quality Rights & Licensing Clarity Navigation & Wayfinding Performance Item Actions & Downloads UI & Visual Design
FigJam affinity mapping session — 10 themes surfaced from 25+ responses.

4. action plan.

every issue was tagged by priority, owner, and evidence from people.

Issues prioritised as CRITICAL / HIGH / MEDIUM, owned by Design / Dev / Metadata.

The prioritised action plan and the structured intake form replacing open board access.

Add on: Before the action plan, anyone with Hive access could add issues directly — duplicates and vague tickets piled up. Along with the PM, I created a structured intake form that replaced open board access, so that every task was reviewed and assigned appropriately.

Redesign:

One of the main issues from the survey was the advanced search page, which was the first design task I set out to tackle.

I went through multiple design iterations, went through feedback loops, used AI as my design partner and finally delivered an interactive HTML prototype to the developers. Figma for visual specs; HTML as the reference for behaviour.

what changed, and why.

01

Sticky submit bar.

  • Submit button was buried below all filters: users scrolled past every filter to reach it.
  • No visibility of result count before clicking: users hit zero-result dead ends and restarted from scratch.
  • New design: Submit pinned to the viewport with a live result count beside it.

"With Advanced Search, it was hard to find the 'submit query' button, all the way at the bottom of the page."

— Survey Respondent

Before / after — interactive prototype.

02

What does "Fuzzy Match" even mean?

Boolean operators and fuzzy/exact dropdowns confused most users. Replaced with plain-language options and an inline tip.

"Wait, what does fuzzy match mean? I have never seen that before."

— Survey respondent

Before / after — interactive prototype.

03

Diacritics and casing shouldn't break search.

'sebah' and 'Sébah' returned different results. Fixed so search works the way people naturally type.

"'sebah' and 'Sébah' return different results. Exact match is case-sensitive unexpectedly."

— Survey respondent

Before / after — interactive prototype.

04

Open Access as a toggle, with context.

Three unlabelled radio buttons → one toggle, with a tooltip explaining what "Open Access" means on Node.

"Is there an explanation on the site for 'Open Access?' Not all users may understand the distinction."

— Survey Respondent

Before / after — interactive prototype.

05

Media Type as icon cards.

Flat checkboxes → visual cards. Scannable, selectable, multiple at once.

Before / after — interactive prototype.

06

Active filters are always visible.

Selections appear as chips in a sticky bar with a live result count.

Before / after — interactive prototype.

07

One decision at a time.

Embedded checkbox lists were easy to accidentally scroll past. Filters now collapse into accordions — one open at a time.

Before / after — interactive prototype.

08

A search guide is also in the works — an inline reference for field options and query patterns.

Accessibility:

Accessibility for UChicagoNode was never an afterthought.

We collaborated with UChicago's Center for Digital Accessibility and reviewed the platform for screen reader compatibility, keyboard navigation, responsiveness and color contrast.

99/100 Siteimprove Score  |  100/100 Level A · AA · AAA · WAI-ARIA  |  84.6 Best Practices

Siteimprove accessibility dashboard showing 99.0 accessibility score and 100/100 on Level A, AA, AAA, and WAI-ARIA
Accessibility review with UChicago's Center for Digital Accessibility.

Learnings:

01.
Structure matters more than timeline. A focused survey outperforms a loose, longer effort.
02.
User quotes and a priority list make it easier to push back. A feeling isn't enough.
03.
The most useful work wasn't the screens, it was getting everyone aligned on the same problems.
04.
HTML prototypes surface different questions than static Figma files. It was interesting how AI made idea iteration and explanation easier.
← BACK NEXT →